Highly active antiretroviral therapy (HAART) has transformed pediatric human immunodeficiency virus (HIV) into a chronic disease. As patients live longer, monitoring long-term response to therapy through the use of laboratory markers (CD4 and viral load) is critical to optimize outcomes; but monitoring increases the cost and complexity of HIV care. Providing HAART to children in developing countries is becoming possible; however, many developing countries do not have the capacity to monitor therapy. Some countries with intermediate resources are already providing HAART -- often without monitoring the effectiveness of care. Although not technically difficult, frequent laboratory testing of HIV RNA viral load and CD4 count is a hidden cost of providing HAART. The long-term goal of this proposal is to provide preliminary data so the principal investigator may conduct prospective international trials using clinical parameters to predict ART failure. This proposal will be an important first step to determine if a combination of clinical parameters (age, gender, growth-velocity, and ethnicity) can predict relevant laboratory surrogate markers, clinical outcomes and ART failure. We will generate a model that predicts one aspect of ART failure (CD4%) in children based on growth velocity using a data set generated in North American HIV-positive children. We will refine the model in a data set in Romanian HIV-positive children; and will test the model prospectively and internationally in a pilot trial in Romania. Following are the specific objectives within this proposal: 1) Develop a clinically based model to predict ART failure to generate immune reconstitution (rise in CD4%) in a retrospective examination of North American HIV positive children, 2) Test and refine the model in a retrospective examination of Romanian HIV positive children, and 3) Assemble a cohort of 50 HIV positive Romanian children; follow them prospectively for 12 months; determine ART failure based on clinical and laboratory progression of HIV while blinded to clinical predictive factors; and test the predictive model. This cohort will provide the proof of concept and the preliminary data to furnish the principal investigator with the ability to apply for large peer-reviewed grants that will allow the model to undergo country-specific testing and modification.